CN108389171A - A kind of light field deblurring and depth estimation method based on Combined estimator fuzzy variable - Google Patents
A kind of light field deblurring and depth estimation method based on Combined estimator fuzzy variable Download PDFInfo
- Publication number
- CN108389171A CN108389171A CN201810189606.2A CN201810189606A CN108389171A CN 108389171 A CN108389171 A CN 108389171A CN 201810189606 A CN201810189606 A CN 201810189606A CN 108389171 A CN108389171 A CN 108389171A
- Authority
- CN
- China
- Prior art keywords
- image
- sub
- light field
- fuzzy
- depth
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
- 238000000034 method Methods 0.000 title claims abstract description 25
- 230000033001 locomotion Effects 0.000 claims abstract description 37
- 238000005457 optimization Methods 0.000 claims abstract description 10
- 239000011159 matrix material Substances 0.000 claims description 12
- 238000005070 sampling Methods 0.000 claims description 9
- 238000013459 approach Methods 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 6
- 230000001419 dependent effect Effects 0.000 claims description 3
- 230000036544 posture Effects 0.000 claims description 3
- 238000011084 recovery Methods 0.000 claims description 3
- 230000000007 visual effect Effects 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 210000000056 organ Anatomy 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 1
- 238000002059 diagnostic imaging Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000000192 social effect Effects 0.000 description 1
Classifications
-
- G06T5/73—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/269—Analysis of motion using gradient-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10052—Images from lightfield camera
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
Abstract
A kind of the light field deblurring and depth estimation method based on Combined estimator fuzzy variable proposed in the present invention, main contents include：Sub-aperture image, light field fuzzy model, the update of sub-image, the update of camera motion and depth map, its process is, reference angular position and timestamp of the median of the centre view and aperture time that first select sub-aperture image as sub-image, use the input sub-aperture image initial depth map of light field, camera motion obscures kernel and the initialization of initial scene depth from local linear, then combined optimization sub-image, depth map and camera motion finally obtain clearly image, light field, depth map and camera motion and are used as output.Combined estimator proposed by the invention realizes the light field deblurring and estimation of Depth of high quality simultaneously under arbitrary 6 degree of freedom camera motion and without constraint scene depth, the clarity for substantially increasing image, the operational excellence in the case where general camera motion and scene depth change.
Description
Technical field
The present invention relates to image restoration fields, and mould is removed more particularly, to a kind of light field based on Combined estimator fuzzy variable
Paste and depth estimation method.
Background technology
In digital age, an important branch of the image deblurring as image restoration technology always is one
Extremely challenging problem has great researching value and social effect.Due to shooting operation or shooting environmental factor etc.
Influence, often there is the phenomenon that fuzzy and distortion in shooting image, or during generation, transmission, record, storage due at
As the reasons such as not perfect of system, transmission medium or recording equipment cause that image information is lost or quality reduces.Therefore image is gone
Fuzzy technology is all particularly important in many fields.Such as in security field, monitoring video equipment record and the picture shot,
When target person or object quickly move, the picture absorbed generally can be relatively fuzzyyer, then may be used by Smear-eliminated technique of image
To generate more clearly image, facilitate the effective information of the acquisitions such as security personnel or police.In Medical Imaging, since human body is each
A organ internal environment is complicated and changeable, therefore the organ internal image shot is often difficult to it is clear that by application image deblurring
Technology can then restore clear image to a certain extent, to help diagnosis and treatment.It is visited in astronomical observation, space
In the fields such as rope, remote sensing prediction, since the field of exploration has non-intellectual, complexity and variability, gone with greater need for image
Fuzzy technology come help researcher restore clear image, so that it is determined that environmental information.However, existing Smear-eliminated technique of image
It is difficult to realize only estimate scene depth with single observation, and still cannot handle completely uneven caused by scene depth variation
It is fuzzy.
The present invention proposes a kind of light field deblurring and depth estimation method based on Combined estimator fuzzy variable, first selects
The reference angular position and timestamp of the centre view of sub-aperture image and the median of aperture time as sub-image, use light field
Sub-aperture image initial depth map is inputted, camera motion obscures kernel and the initialization of initial scene depth from local linear, connects
Combined optimization sub-image, depth map and camera motion, clearly image, light field, depth map and camera motion conduct are finally obtained
Output.Combined estimator proposed by the invention realizes simultaneously under arbitrary 6- degree of freedom camera motion and without constraint scene depth
The light field deblurring and estimation of Depth of high quality, substantially increase the clarity of image, in general camera motion and scene depth
Operational excellence in the case of variation.
Invention content
The problems such as being difficult to only estimate scene depth with single observation, the purpose of the present invention is to provide one kind based on connection
The light field deblurring and depth estimation method of ambiguous estimation variable are closed, the centre view and aperture time of sub-aperture image are first selected
Reference angular position and timestamp of the median as sub-image, use the input sub-aperture image initial depth map of light field, phase
Machine movement obscures kernel and the initialization of initial scene depth from local linear, then combined optimization sub-image, depth map and camera fortune
It is dynamic, it finally obtains clearly image, light field, depth map and camera motion and is used as output.
Estimate to solve the above problems, the present invention provides a kind of light field deblurring and depth based on Combined estimator fuzzy variable
Meter method, main contents include：
(1) sub-aperture image；
(2) light field fuzzy model；
(3) update of sub-image；
(4) update of camera motion and depth map.
Wherein, the sub-aperture image, there are four coordinates for the pixel tool in four-dimensional light field, that is, are used for (x, y) in space
With (u, v) for angular coordinate；Light field, which can be considered as one group of u × v, has the multi-view image of narrow baseline, commonly referred to as sub-aperture
Diameter image I (x, u)；Wherein x=(x, y) and u=(u, v)；For each sub-aperture image, blurred picture B (x, u) is in shutter
Clear image I when openingtAverage value [the t of (x, u)0,t1]；By projecting the single sub-image moved with Three-dimensional Rigidity come close
Like all fuzzy sub-aperture images.
Further, the approximation selects the median (t of the centre view (c) and aperture time of sub-aperture imager)
Reference angular position and timestamp as sub-image；Then from each sub-aperture image to sub-imagePixel correspondence expression
It is as follows：
wt(x, u) is calculated from u to c and from t to trWarpage location of pixels；MatrixWithIndicate respective corners
Spend the 6- degree of freedom camera postures and timestamp at position；Dt(x, u) is the depth map at timestamp t；
In the model proposed, fuzzy operator Ψ () by by the integral approach of B (x, u) for following finite sum come
Definition：
In formula (2), tmIt is interim [t0,t1] m-th of uniform sampling timestamp.
Further, the finite sum determines there was only central viewpoint variableI.e.With WithIt is the variable related with u in warping function (2)；Therefore, by using
Centre view variable parameterizesWithDue to relative pose Pc→uTime to time change, so
Wherein, exp and log indicates the index and logarithmic mapping between Lie group SE (3) and Lie algebra se (3) space；In order to
The viewpoint offsets of sub-image are reduced to the maximum extent, it is assumed thatWork as tm=trWhen so thatAs unit matrix；Also byIt is indicated to warpage and interpolation by preceding.
Wherein, the light field fuzzy model, in order to estimate all fuzzy variables in proposed light field fuzzy model,
It needs to restore latent variable, i.e.,WithEnergy function is modeled as follows：
Data item forces the fuzzy brightness uniformity between light field and the light field of recovery of input；Last two are creep respectively
The total variation regularization of amount and depth map.
Further, energy function modeling, in energy model,WithImplicitly it is included in warpage letter
In number (2)；Optimize three latent variables in an alternating fashion；A variable is minimized, and its dependent variable is fixed；Needle successively
To three variable optimization formula (4)；Carry out approximate L1 using iteration again weighted least-squares method (IRLS) to optimize；Optimization process with
A small amount of iteration convergence (<10).
Wherein, the update of the sub-image, the algorithm update sub-image firstIn data item, if
WithIt remains unchanged, then fuzzy operator (2) is reduced to linear matrix multiplication；Update sub-image is equal to the following institute of minimum (4)
Show：
It is vectorial images, andIt is the fuzzy operator of square matrix form, wherein n is
The quantity of pixel in centre visual angle sub-aperture image；Total variation regularization is eliminated as prior to the sub-image with clear boundary
Artifact.
Further, the update of the camera motion and depth map, since formula (2) isWithIt is non-thread
Property function needs to approach it in linear form to efficiently calculate；Fuzzy operation (2) is approximately single order extension；Enable D0(x,c)
WithIndicate initializaing variable, then formula (2) is approximately as described below：
F is the movement stream generated by warping function, andIndicate the six-vector on se (3)；
Once usingWithApproximation, formula (4) can be optimized using IRLS；It obtains
WithIt is current respectivelyWithIncrement size；They update as follows：
Wherein,Pass through motion vectorIndex mapping update.
Further, the camera motion uses the input sub-aperture image initial depth map of light field first；It is false
Determine camera not move and minimize formula (4) to obtain initiallyIt is simple more as one to minimize formula (4)
View stereo matching problem；Camera motionKernel and the initialization of initial scene depth are obscured from local linear.
Further, depth initialization, first using the local linear fuzzy core of estimation B (x, c)；Then, intend
The pixel coordinate moved by linear kernel and the coordinate projected again by warping function are closed, as follows：
Wherein, xiIt is the location of pixels of sampling, l (xi) it is xiThe point moved by the terminal of linear kernel；By fitting byThe mobile x with l ()iTo obtainSince scene depth is fixed on initial depth figure,It isOnly
One variable；Random sampling unification algorism (RANSAC) is for searching the camera motion that can most describe the linear kernel of pixel；N is random
The quantity of sample is always 4.
Description of the drawings
Fig. 1 is a kind of system stream of light field deblurring and depth estimation method based on Combined estimator fuzzy variable of the present invention
Cheng Tu.
Fig. 2 is that the present invention a kind of light field deblurring and combining for depth estimation method based on Combined estimator fuzzy variable are estimated
Meter.
Fig. 3 is the iteration connection of the present invention a kind of light field deblurring and depth estimation method based on Combined estimator fuzzy variable
Close estimation example.
Specific implementation mode
It should be noted that in the absence of conflict, the features in the embodiments and the embodiments of the present application can phase
It mutually combines, invention is further described in detail in the following with reference to the drawings and specific embodiments.
Fig. 1 is a kind of system stream of light field deblurring and depth estimation method based on Combined estimator fuzzy variable of the present invention
Cheng Tu.Include mainly sub-aperture image, light field fuzzy model, the update of sub-image, the update of camera motion and depth map.
Main process is：Reference of the median of the centre view and aperture time that first select sub-aperture image as sub-image
Angle Position and timestamp, using the input sub-aperture image initial depth map of light field, camera motion is fuzzy interior from local linear
Core and the initialization of initial scene depth, then combined optimization sub-image, depth map and camera motion, finally obtain clearly image,
Light field, depth map and camera motion are as output.
Sub-aperture image, there are four coordinates for the pixel tool in four-dimensional light field, that is, are used for (x, y) in space and are used for angular coordinate
(u, v)；Light field can be considered as one group of u × v have narrow baseline multi-view image, commonly referred to as sub-aperture image I (x,
u)；Wherein x=(x, y) and u=(u, v)；For each sub-aperture image, blurred picture B (x, u) is clear when shutter is opened
Clear image ItAverage value [the t of (x, u)0,t1]；By projecting the single sub-image moved with Three-dimensional Rigidity come approximate all fuzzy
Sub-aperture image.
Select the median (t of the centre view (c) and aperture time of sub-aperture imager) reference angular position as sub-image
And timestamp；Then from each sub-aperture image to sub-imagePixel correspondence expression it is as follows：
wt(x, u) is calculated from u to c and from t to trWarpage location of pixels；MatrixWithIndicate respective corners
Spend the 6- degree of freedom camera postures and timestamp at position；Dt(x, u) is the depth map at timestamp t；
In the model proposed, fuzzy operator Ψ () by by the integral approach of B (x, u) for following finite sum come
Definition：
In formula (2), tmIt is interim [t0,t1] m-th of uniform sampling timestamp.
Determine there was only central viewpoint variableI.e.With With
It is the variable related with u in warping function (2)；Therefore, it is parameterized by using centre view variableWith
Due to relative pose Pc→uTime to time change, so
Wherein, exp and log indicates the index and logarithmic mapping between Lie group SE (3) and Lie algebra se (3) space；In order to
The viewpoint offsets of sub-image are reduced to the maximum extent, it is assumed thatWork as tm=trWhen so thatAs unit matrix；Also byIt is indicated to warpage and interpolation by preceding.
Light field fuzzy model needs to restore latent to estimate all fuzzy variables in proposed light field fuzzy model
In variable, i.e.,WithEnergy function is modeled as follows：
Data item forces the fuzzy brightness uniformity between light field and the light field of recovery of input；Last two are creep respectively
The total variation regularization of amount and depth map.
In energy model,WithImplicitly it is included in warping function (2)；Optimize three in an alternating fashion
Latent variable；A variable is minimized, and its dependent variable is fixed；It is directed to three variables successively and optimizes formula (4)；Using repeatedly
In generation, weighted least-squares method (IRLS) carried out approximate L1 optimizations again；Optimization process with a small amount of iteration convergence (<10).
Fig. 2 is that the present invention a kind of light field deblurring and combining for depth estimation method based on Combined estimator fuzzy variable are estimated
Meter.The algorithm from single light field Combined estimator sub-image depth map and camera motion.Figure (a) is fuzzy light field sub-aperture image
Centre view；Figure (b) is the blurred picture for scheming (a)；Figure (c) is the depth map of estimation；It is camera motion path and direction to scheme (d)
(6- degree of freedom).
The algorithm updates sub-image firstIn data item, ifWithIt remains unchanged, then obscures and calculate
Sub (2) are reduced to linear matrix multiplication；It is as follows that update sub-image is equal to minimum (4)：
It is vectorial images, andIt is the fuzzy operator of square matrix form, wherein n is
The quantity of pixel in centre visual angle sub-aperture image；Total variation regularization is eliminated as prior to the sub-image with clear boundary
Artifact.
Since formula (2) isWithNonlinear function need to approach in linear form to efficiently calculate
It；Fuzzy operation (2) is approximately single order extension；Enable D0(x, c) andIndicate initializaing variable, then formula (2) is approximately as described below：
F is the movement stream generated by warping function, andIndicate the six-vector on se (3)；
Once usingWithApproximation, formula (4) can be optimized using IRLS；It obtains
WithIt is current respectivelyWithIncrement size；They update as follows：
Wherein,Pass through motion vectorIndex mapping update.
First, using the input sub-aperture image initial depth map of light field；It is assumed that camera does not move and minimizes public affairs
Formula (4) is initial to obtainMinimizing formula (4) becomes a simple multiple view stereo matching problem；Camera motionKernel and the initialization of initial scene depth are obscured from local linear.
First using the local linear fuzzy core of estimation B (x, c)；Then, fitting is moved by linear kernel pixel coordinate and
The coordinate projected again by warping function, as follows：
Wherein, xiIt is the location of pixels of sampling, l (xi) it is xiThe point moved by the terminal of linear kernel；By fitting byThe mobile x with l ()iTo obtainSince scene depth is fixed on initial depth figure,It isOnly
One variable；Random sampling unification algorism (RANSAC) is for searching the camera motion that can most describe the linear kernel of pixel；N is random
The quantity of sample is always 4.
Fig. 3 is the iteration connection of the present invention a kind of light field deblurring and depth estimation method based on Combined estimator fuzzy variable
Close estimation example.The method proposed restrains in a small amount of iteration.Figure (a) and figure (b) input blurred picture by iteration and go
Fuzzy result.Scheme (c) and figure (d) initially obscures depth map and depth estimation result by iteration.Joint proposed by the invention
Estimate in arbitrary 6- degree of freedom camera motion and without the light field deblurring and depth for constraining under scene depth while realizing high quality
Degree estimation, substantially increases the clarity of image, the operational excellence in the case where general camera motion and scene depth change.
For those skilled in the art, the present invention is not limited to the details of above-described embodiment, in the essence without departing substantially from the present invention
In the case of refreshing and range, the present invention can be realized in other specific forms.In addition, those skilled in the art can be to this hair
Bright to carry out various modification and variations without departing from the spirit and scope of the present invention, these improvements and modifications also should be regarded as the present invention's
Protection domain.Therefore, the following claims are intended to be interpreted as including preferred embodiment and falls into all changes of the scope of the invention
More and change.
Claims (10)
1. a kind of light field deblurring and depth estimation method based on Combined estimator fuzzy variable, which is characterized in that include mainly
Sub-aperture image (one)；Light field fuzzy model (two)；The update (three) of sub-image；The update (four) of camera motion and depth map.
2. based on the sub-aperture image (one) described in claims 1, which is characterized in that there are four the pixel tools in four-dimensional light field
Coordinate is used for (x, y) in space and (u, v) for angular coordinate；Light field, which can be considered as one group of u × v, has narrow baseline
Multi-view image, commonly referred to as sub-aperture image I (x, u)；Wherein x=(x, y) and u=(u, v)；For each sub-aperture figure
Picture, blurred picture B (x, u) are the clear image I when shutter is openedtAverage value [the t of (x, u)0,t1]；Have three by projection
The single sub-image for tieing up rigid motion carrys out approximate all fuzzy sub-aperture images.
3. based on the approximation described in claims 2, which is characterized in that the centre view (c) and shutter of selection sub-aperture image
Median (the t of timer) reference angular position and timestamp as sub-image；Then from each sub-aperture image to sub-image's
The expression of pixel correspondence is as follows：
wt(x, u) is calculated from u to c and from t to trWarpage location of pixels；MatrixWithIndicate respective angles position
Set the 6- degree of freedom camera postures and timestamp at place；Dt(x, u) is the depth map at timestamp t；
In the model proposed, fuzzy operator Ψ () is by determining the integral approach of B (x, u) for following finite sum
Justice：
In formula (2), tmIt is interim [t0,t1] m-th of uniform sampling timestamp.
4. based on the finite sum described in claims 3, which is characterized in that determine there was only central viewpoint variableI.e.With WithIt is the variable related with u in warping function (2)；
Therefore, it is parameterized by using centre view variableWithDue to relative pose Pc→uTime to time change, institute
With
Wherein, exp and log indicates Lie group SE (3) and Lie algebraIndex between space and logarithmic mapping；For maximum limit
Reduce the viewpoint offsets of sub-image in degree ground, it is assumed thatWork as tm=trWhen so thatAs unit matrix；
Also byIt is indicated to warpage and interpolation by preceding.
5. based on the light field fuzzy model (two) described in claims 1, which is characterized in that in order to estimate proposed light field mould
All fuzzy variables in fuzzy model need to restore latent variable, i.e.,WithEnergy function is modeled
It is as follows：
Data item forces the fuzzy brightness uniformity between light field and the light field of recovery of input；Last two be respectively latent variable and
The total variation regularization of depth map.
6. based on the energy function modeling described in claims 5, which is characterized in that in energy model,WithIt is hidden
Formula is included in warping function (2)；Optimize three latent variables in an alternating fashion；One variable of minimum, and its dependent variable
It is fixed；It is directed to three variables successively and optimizes formula (4)；Carry out approximate L1 using iteration again weighted least-squares method (IRLS)
Optimization；Optimization process with a small amount of iteration convergence (<10).
7. the update (three) based on the sub-image described in claims 1, which is characterized in that the algorithm updates sub-image firstIn data item, ifWithIt remains unchanged, then fuzzy operator (2) is reduced to linear matrix multiplication；
It is as follows that update sub-image is equal to minimum (4)：
It is vectorial images, andIt is the fuzzy operator of square matrix form, wherein n is center
The quantity of pixel in the sub-aperture image of visual angle；Total variation regularization eliminates puppet as prior to the sub-image with clear boundary
Shadow.
8. the update (four) based on camera motion and depth map described in claims 1, which is characterized in that due to formula (2)
It isWithNonlinear function need to approach it in linear form to efficiently calculate；Fuzzy operation (2) is approximately
Single order extends；Enable D0(x, c) andIndicate initializaing variable, then formula (2) is approximately as described below：
F is the movement stream generated by warping function, andIt indicatesOn six-vector；
Once usingWithApproximation, formula (4) can be optimized using IRLS；It obtainsWithPoint
It is not currentWithIncrement size；They update as follows：
Wherein,Pass through motion vectorIndex mapping update.
9. based on the camera motion described in claims 8, which is characterized in that first, use the input sub-aperture image of light field
Initialize depth map；It is assumed that camera does not move and minimizes formula (4) to obtain initiallyMinimize formula (4) at
For a simple multiple view stereo matching problem；Camera motionKernel is obscured from local linear and initial scene depth is initial
Change.
10. based on the depth initialization described in claims 9, which is characterized in that first using the local line of estimation B (x, c)
Property fuzzy core；Then, the pixel coordinate moved by linear kernel and the coordinate projected again by warping function, following institute are fitted
Show：
Wherein, xiIt is the location of pixels of sampling, l (xi) it is xiThe point moved by the terminal of linear kernel；By fitting by
The mobile x with l ()iTo obtainSince scene depth is fixed on initial depth figure,It isUnique variable；
Random sampling unification algorism (RANSAC) is for searching the camera motion that can most describe the linear kernel of pixel；N is random sample
Quantity is always 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810189606.2A CN108389171A (en) | 2018-03-08 | 2018-03-08 | A kind of light field deblurring and depth estimation method based on Combined estimator fuzzy variable |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810189606.2A CN108389171A (en) | 2018-03-08 | 2018-03-08 | A kind of light field deblurring and depth estimation method based on Combined estimator fuzzy variable |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108389171A true CN108389171A (en) | 2018-08-10 |
Family
ID=63067024
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810189606.2A Withdrawn CN108389171A (en) | 2018-03-08 | 2018-03-08 | A kind of light field deblurring and depth estimation method based on Combined estimator fuzzy variable |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108389171A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110706346A (en) * | 2019-09-17 | 2020-01-17 | 北京优科核动科技发展有限公司 | Space-time joint optimization reconstruction method and system |
| CN111179333A (en) * | 2019-12-09 | 2020-05-19 | 天津大学 | Defocus fuzzy kernel estimation method based on binocular stereo vision |
| CN111191618A (en) * | 2020-01-02 | 2020-05-22 | 武汉大学 | KNN scene classification method and system based on matrix group |
| CN112150526A (en) * | 2020-07-27 | 2020-12-29 | 浙江大学 | Light field image depth estimation method based on depth learning |
| CN112184731A (en) * | 2020-09-28 | 2021-01-05 | 北京工业大学 | Multi-view stereo depth estimation method based on antagonism training |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105704371A (en) * | 2016-01-25 | 2016-06-22 | 深圳市未来媒体技术研究院 | Light field refocusing method |
| CN106803892A (en) * | 2017-03-13 | 2017-06-06 | 中国科学院光电技术研究所 | A kind of light field high-resolution imaging method based on Optical field measurement |
-
2018
- 2018-03-08 CN CN201810189606.2A patent/CN108389171A/en not_active Withdrawn
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105704371A (en) * | 2016-01-25 | 2016-06-22 | 深圳市未来媒体技术研究院 | Light field refocusing method |
| CN106803892A (en) * | 2017-03-13 | 2017-06-06 | 中国科学院光电技术研究所 | A kind of light field high-resolution imaging method based on Optical field measurement |
Non-Patent Citations (1)
| Title |
|---|
| DONGWOO LEE: "Joint Blind Motion Deblurring and Depth Estimation of Light Field", 《ARXIV:1711.10918V1》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110706346A (en) * | 2019-09-17 | 2020-01-17 | 北京优科核动科技发展有限公司 | Space-time joint optimization reconstruction method and system |
| CN110706346B (en) * | 2019-09-17 | 2022-11-15 | 浙江荷湖科技有限公司 | Space-time joint optimization reconstruction method and system |
| CN111179333A (en) * | 2019-12-09 | 2020-05-19 | 天津大学 | Defocus fuzzy kernel estimation method based on binocular stereo vision |
| CN111179333B (en) * | 2019-12-09 | 2024-04-26 | 天津大学 | Defocus blur kernel estimation method based on binocular stereo vision |
| CN111191618A (en) * | 2020-01-02 | 2020-05-22 | 武汉大学 | KNN scene classification method and system based on matrix group |
| CN112150526A (en) * | 2020-07-27 | 2020-12-29 | 浙江大学 | Light field image depth estimation method based on depth learning |
| CN112184731A (en) * | 2020-09-28 | 2021-01-05 | 北京工业大学 | Multi-view stereo depth estimation method based on antagonism training |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108389171A (en) | A kind of light field deblurring and depth estimation method based on Combined estimator fuzzy variable | |
| Reinbacher et al. | Real-time panoramic tracking for event cameras | |
| Gallego et al. | Accurate angular velocity estimation with an event camera | |
| JP6563609B2 (en) | Efficient canvas view generation from intermediate views | |
| Im et al. | All-around depth from small motion with a spherical panoramic camera | |
| US10334168B2 (en) | Threshold determination in a RANSAC algorithm | |
| Chin et al. | Star tracking using an event camera | |
| CN110945565A (en) | Dense visual SLAM using probabilistic bin maps | |
| WO2012083982A1 (en) | Method for determining a parameter set designed for determining the pose of a camera and/or for determining a three-dimensional structure of the at least one real object | |
| CN105069753B (en) | A kind of shake Restoration method of blurred image of facing moving terminal | |
| US11887256B2 (en) | Deferred neural rendering for view extrapolation | |
| US11049313B2 (en) | Rendering an object | |
| Kim et al. | Real-time panorama canvas of natural images | |
| Dellaert et al. | Super-resolved texture tracking of planar surface patches | |
| CN111192308B (en) | Image processing method and device, electronic equipment and computer storage medium | |
| Xian et al. | Neural Lens Modeling | |
| Peng et al. | PDRF: progressively deblurring radiance field for fast scene reconstruction from blurry images | |
| JP6341540B2 (en) | Information terminal device, method and program | |
| Gilbert et al. | Inpainting of wide-baseline multiple viewpoint video | |
| Cao et al. | Make object connect: A pose estimation network for UAV images of the outdoor scene | |
| Lin et al. | Reinforcement learning-based image exposure reconstruction for homography estimation | |
| KR102298098B1 (en) | Method and Apparatus for Generating 3D Model through Tracking of RGB-D Camera | |
| Lee et al. | High dynamic range imaging via truncated nuclear norm minimization of low-rank matrix | |
| Guo et al. | Single Image based Fog Information Estimation for Virtual Objects in A Foggy Scene | |
| Toklu et al. | 2-D mesh-based synthetic transfiguration of an object with occlusion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20180810 |
|
| WW01 | Invention patent application withdrawn after publication |